RNA-seq basics: From reads to differential expression COMBINE - - PowerPoint PPT Presentation

▶
rna seq basics from reads to differential expression
SMART_READER_LITE
LIVE PREVIEW

RNA-seq basics: From reads to differential expression COMBINE - - PowerPoint PPT Presentation

Feb 09, 2024 386 likes •671 views

RNA-seq basics: From reads to differential expression COMBINE RNA-seq Workshop RNA sequencing (RNA-seq) Use of ultra high-throughput sequencing (next- or second-generation) technologies to study gene expression Many

slide-1
SLIDE 1

RNA-seq basics: From reads to differential expression

COMBINE RNA-seq Workshop

slide-2
SLIDE 2

RNA sequencing (RNA-seq)

  • Use of ultra high-throughput sequencing (‘next-’ or

‘second’-generation) technologies to study gene expression

  • Many applications: differential expression,

transcript discovery, splice variants, allele-specific expression

  • In this hands-on course, you will learn how to use

statistical methods to assess differential expression in RNA-seq data using popular tools in R/Bioconductor

slide-3
SLIDE 3

Genes and transcripts

Gene transcript

Slide from Alicia Oshlack

slide-4
SLIDE 4

From transcripts to short reads

Pepke et al, Nature Methods, 2009

slide-5
SLIDE 5

Raw data comes in fastq files

  • Short sequence reads
  • Quality scores

@HWI-ST1148:308:C694RACXX:5:1101:1768:1990 1:N:0:CGTACG NTAGGCCTTGGCAGTTTTGGAGAATCACTGCTGCCAAAGAGTCTACTTGG + #0<FFFFFFFFFFIIIIIIIIIIIIIIIIIIIIIIIIIIIIFFIIIIIII @HWI-ST1148:308:C694RACXX:5:1101:3409:1990 1:N:0:CGTACG NAGTTACCCTAGGGATAACAGCGCAATCCTATTCTAGAGTCCATATCAAC + #000BFBFFFFFFF<BFFFFBBBBBFBBFF<<FBFFIBFFFBFFFIIBFF 50 bp sequence

slide-6
SLIDE 6

RNA-seq analysis steps

Raw sequence reads Map to genome Summarize reads over genes Statistical testing: Determine differentially expressed genes Pathway analysis

Slide from Alicia Oshlack

slide-7
SLIDE 7

Mapping reads to the genome

  • Where do the millions of short sequences come

from in the genome?

  • Sequencing transcripts, not the genome

CDS CDS CDS CDS CDS CDS CDS CDS

Gene transcript

slide-8
SLIDE 8

Lots of good aligners handle splice junctions well

Exon 1 Exon 2

slide-9
SLIDE 9

Aligned reads (bam files)

HWI-ST1148:308:C694RACXX:6:2209:15171:26188 272 chr10 76314 0 50M * 0 0 CATCTGATCTTTGACAAACCTGACAAACACAAGCAATGGGGAAAGGATTC IIIIIIIIIIIIIIIIIFIFFFIIIIIIIIIIIIIIIFFFFFFFFFFBBB NH:i:10 HI:i:6 AS:i:49 nM:i:0 HWI-ST1148:308:C694RACXX:6:2306:17518:85846 272 chr10 76315 0 50M * 0 0 ATCTGATCTTTGACAAACCTGACAAACACAAGCAATGGGGAAAGGATTCC BFIIIIIIFFIFFBFFFFFFFFFFFFFIIIFFFFFFIFFFFFFFFFFBBB NH:i:10 HI:i:7 AS:i:49 nM:i:0

A row for each sequence Millions of rows...

slide-10
SLIDE 10

RNA-seq analysis steps

Raw sequence reads Map to genome Summarize reads over genes Statistical testing: Determine differentially expressed genes Pathway analysis

Slide from Alicia Oshlack

slide-11
SLIDE 11

Counting over exons vs counting over genes

Exon 1 Exon 2

Exon 1 = 8 reads Exon 2 = 10 reads Counting over whole gene (Exon1 + Exon2) = 15

slide-12
SLIDE 12

Summarization turns mapped reads into a table of counts

Tag ID A1 A2 B1 B2

ENSG00000124208 478 619 4830 7165 ENSG00000182463 27 20 48 55 ENSG00000125835 132 200 560 408 ENSG00000125834 42 60 131 99 ENSG00000197818 21 29 52 44 ENSG00000125831 ENSG00000215443 4 4 9 7 ENSG00000222008 30 23 ENSG00000101444 46 63 54 53 ENSG00000101333 2256 2793 2702 2976

… … tens of thousands more tags …

** very high dimensional data **

slide-13
SLIDE 13

RNA-seq analysis steps

Raw sequence reads Map to genome Summarize reads over genes Statistical testing: Determine differentially expressed genes Pathway analysis

Slide from Alicia Oshlack

slide-14
SLIDE 14

Assessing differential expression

  • For each gene in each sample we have a

measure of abundance

– Number of reads mapping across gene

  • We want to know whether there is a

statistically significant difference in abundance between treatments/groups/genotypes

slide-15
SLIDE 15

Is this gene differentially expressed?

Group 1 Group 2

Data from Shireen Lamande

slide-16
SLIDE 16

Group 1 Group 2

  • utlier

Replication is really important!

Is this gene differentially expressed?

slide-17
SLIDE 17

Quality control – check your data!

Data from Andrew Elefanty Sorted cell populations

slide-18
SLIDE 18

Things to think about before statistical testing

  • Filtering out lowly expressed genes

– Need to make decisions about cut-offs – Can be an iterative process

Want to avoid calling this gene DE due to one sample

slide-19
SLIDE 19
  • Normalisation

– Library size (sequencing depth) – Composition bias (TMM) – Batch effects (RUVSeq)

Things to think about before statistical testing

.

  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ● ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • ● ●
  • ●
  • ●
  • ●
  • ●
  • ●
  • 20
  • 15
  • 10
  • 5

5 A = log2( Liver NL Kidney NK) M = log2(Liver NL) - log2(Kidney NK)

  • ●
  • Housekeeping genes

Unique to a sample

(c)

slide-20
SLIDE 20

Statistical testing for DE

  • For each gene, is the mean expression level

under one condition significantly different from the mean expression level under a different condition?

Tag ID A1 A2 B1 B2

ENSG00000124208 478 619 4830 7165 ENSG00000182463 27 20 48 55 ENSG00000125835 132 200 560 408 ENSG00000125834 42 60 131 99

… … tens of thousands more tags … 20

slide-21
SLIDE 21

Many different statistical methods

  • Model the counts directly

– Negative binomial modelling is best because it captures biological as well as technical variability – Most popular packages in R

  • edgeR
  • DESeq/DESeq2
  • Lots of others exist (baySeq, NBPSeq, …)
  • Transform the counts and used normal based

methods

– voom in the limma package

slide-22
SLIDE 22

Statistical testing gives each gene a p-value for evidence of DE

Tag ID A1 A2 B1 B2

ENSG00000124208 478 619 4830 7165 ENSG00000182463 27 20 48 55 ENSG00000125835 132 200 560 408 ENSG00000125834 42 60 131 99 ENSG00000197818 21 29 52 44 ENSG00000125831 ENSG00000215443 4 4 9 7 ENSG00000222008 30 23 ENSG00000101444 46 63 54 53 ENSG00000101333 2256 2793 2702 2976

… … tens of thousands more tags …

Tag ID P-value

ENSG00000124208 0.0002 ENSG00000182463 0.12 ENSG00000125835 0.034 ENSG00000125834 0.08 ENSG00000197818 0.64 ENSG00000125831 1 ENSG00000215443 1 ENSG00000222008 0.06 ENSG00000101444 0.73 ENSG00000101333 0.22

…

slide-23
SLIDE 23

RNA-seq analysis steps

Raw sequence reads Map to genome Summarize reads over genes Statistical testing: Determine differentially expressed genes Pathway analysis

Slide from Alicia Oshlack

Learn something!

slide-24
SLIDE 24

Summary

  • Lots of choices in analysis methodology
  • Quality control is essential! Sometimes

detective work is necessary.

  • Each step of the analysis requires decisions

that impact down-stream analysis

  • Life gets harder when there’s no genome or

poor quality genomes

slide-25
SLIDE 25

RNA-seq analysis in R / Bioconductor

slide-26
SLIDE 26

Acknowledgements

Slides:

  • Alicia Oshlack
  • Belinda Phipson
  • Anthony Hawkins
  • Gordon Smyth
  • Davis McCarthy

Data:

  • Andrew Elefanty and Elizabeth Ng
  • Shireen Lamande